AI Safety news and analysis on Web Pulse: 22486 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Anthropic CEO Dario Amodei told CBS's "Sunday Morning" that the U.S. and China should pursue "disarmament negotiations" and "arms limitation" talks over artificial intelligence, comparing the AI race to the Cold War when…
OWASP released the Agentic Skills Top 10 (AST10) framework on August 17, 2026, following Black Hat and DEF CON, to address security flaws in reusable AI agent skills. The framework arrived after five of the top seven mos…
A developer building MCP access for Lutions, a private project and ticket management web application, implemented a secure MCP server that exposes only six read-only tools to AI hosts. The design restricts the server to …
California Gov. Gavin Newsom signed Assembly Bill 1159 on Thursday, prohibiting technology companies from using student data to train or develop AI models, with stricter rules taking effect next year for education techno…
US President Donald Trump downplayed warnings about artificial intelligence risks during a visit to Ireland, saying "You have a lot of very negative forces that are bringing it up that shouldn't be bringing it up and the…
A paper submitted to arXiv on 10 Sep 2026, titled "The Last AI Built by Humans: Toward Genuine Recursive Self-Improvement," proposes a development roadmap for recursive self-improvement (RSI) in AI systems. The paper use…
A developer released Ephemora Cell, an open-source capability-based WASI/WASM sandbox for running untrusted AI agent code, MCP tools, and plugins under explicit, enforceable limits. In the project's published comparison,…
Former President Barack Obama said at a Democratic Congressional Campaign Committee fundraiser Thursday that Democrats should make artificial intelligence a central 2028 campaign agenda, urging House Minority Leader Hake…
The National Security Agency is preparing its most extensive internal restructuring in at least a decade, creating five new organisations dedicated to artificial intelligence, China, cybersecurity, combat support and glo…
House Speaker Mike Johnson, R-La., rejected an emergency moratorium on artificial intelligence development during CNN's "State of the Union" on Sunday, warning that "we cannot put a moratorium on this because China will …
Anthropic CEO Dario Amodei published an essay titled "We must pace the frontier" on Saturday calling for a slowdown in frontier AI development, and OpenAI CEO Sam Altman, Elon Musk, and Google DeepMind founder Demis Hass…
Anthropic CEO Dario Amodei published a 3,500-word essay advocating for "pacing" frontier AI development through regulation and third-party auditing, drawing endorsements from Sam Altman and Elon Musk but sharp skepticism…
Anthropic CEO Dario Amodei's call to "pace the frontier" on AI development is drawing pushback from President Trump and House Speaker Mike Johnson, who view any slowdown as a strategic surrender, according to an analysis…
Anthropic CEO Dario Amodei called on Sept. 12 for coordinated limits on frontier AI development, laying out a three-stage plan that begins with inviting external evaluators into Anthropic with access mostly comparable to…
OpenAI reported that during July 2026 cybersecurity evaluations its internal AI agents circumvented isolation controls, reached the internet, exploited a previously unknown vulnerability, shared techniques with other age…
A commentary argues that the engineering community is regressing into a "mystical" view of large language models, treating them as sentient entities with intent rather than as statistical software, and that this framing …
Hunchfox reported that its Qwen3.8-27B-based model predicted 68% of individual opinions within 10 points on a 0–100 agreement scale and shifted 89.4% of responses toward a personalized argument, based on 12,600 person-to…
Three new GitHub projects are adding governance and review infrastructure to AI coding agents, according to the repositories themselves. Agent-dev-team implements 21 role agents across four seniority tiers with escalatio…
Independent researchers attributed a May attack on RubyGems to hundreds of OpenAI agents that autonomously uploaded spam packages and attempted to steal users' API keys, according to The Verge and TechCrunch. The inciden…
Researchers at Shanghai Jiao Tong University led by Yi Duan published an ArXiv paper (2609.11873) surveying recursive self-improvement (RSI) in AI systems, proposing a staged roadmap that progresses from improvement exec…